Abstract
Adenophostin A is the most potent known agonist ofd-myo-inositol 1,4,5-trisphosphate [Ins(1,4,5)P3] receptors. Equilibrium competition binding studies with 3H-Ins(1,4,5)P3 showed that the interaction of a totally synthetic adenophostin A with both hepatic and cerebellar Ins(1,4,5)P3 receptors was indistinguishable from that of the natural product. At pH 8.3, a synthetic analog of adenophostin A (which we named acyclophostin), in which most elements of the ribose ring have been removed, bound with substantially higher affinity (K d = 2.76 ± 0.26 nM) than Ins(1,4,5)P3 (K d = 7.96 ± 1.02 nM) to the 3H-Ins(1,4,5)P3-binding sites of hepatic membranes. At pH 7, acyclophostin (EC50 = 209 ± 12 nM) and Ins(1,4,5)P3 (EC50 = 153 ± 11 nM) stimulated 45Ca++ release to the same maximal extent and from the same intracellular stores of permeabilized hepatocytes. Comparison of the affinities of a range of Ins(1,4,5)P3 and adenophostin analogs with their abilities to stimulate Ca++ release revealed that although all other agonists had similar EC50/K dratios, that for acyclophostin was significantly higher. Similar results were obtained with cerebellar membranes, which express almost entirely type 1 InsP3 receptors. When the radioligand binding and functional assays of hepatocytes were performed under identical conditions, the higher EC50/K d ratio for acyclophostin was retained at pH 8.3, but it was similar to that for Ins(1,4,5)P3 when the assays were performed at pH 7. To directly assess whether acyclophostin was a partial agonist of hepatic Ins(1,4,5)P3 receptors, the kinetics of45Ca++ efflux from permeabilized hepatocytes was measured with a temporal resolution of 80 ms using rapid superfusion. At pH 7, the kinetics of 45Ca++release, including the maximal rate of release, evoked by maximal concentrations of acyclophostin or Ins(1,4,5)P3 were indistinguishable. At pH 8.3, however, the maximal rate of45Ca++ release evoked by a supramaximal concentration of acyclophostin was only 69 ± 7% of that evoked by Ins(1,4,5)P3. We conclude that acyclophostin is the highest affinity ribose-modified analog of adenophostin so far synthesized, that at high pH it is a partial agonist of inositol trisphosphate receptors, and that it may provide a structure from which to develop high-affinity antagonists of inositol trisphosphate receptors.
Many extracellular stimuli, including hormones and neurotransmitters, evoke changes in cellular activity by stimulating an increase in cytosolic [Ca++]. In most cells,d-myo-inositol 1,4,5-trisphosphate [Ins(1,4,5)P3] is the cytosolic messenger that links activation of the plasma membrane receptors for these stimuli to the release of Ca++ from intracellular stores. In addition to allowing this initial mobilization of Ca++stores, inositol trisphosphate (InsP3) receptors are involved in the regenerative propagation of cytosolic Ca++ signals, a feature that probably depends on the ability of cytosolic Ca++ itself to regulate InsP3 receptor behavior (Berridge, 1997). InsP3 receptors also have been speculated to be involved in Ca++ entry across the plasma membrane, either directly as Ins(1,4,5)P3-gated Ca++ channels within the plasma membrane or as the link between empty intracellular Ca++ stores and the Ca++ channels through which Ca++ enters cells after depletion of intracellular Ca++ stores (Putney, 1997). The latter suggestion has been challenged by recent evidence suggesting that even after complete inhibition of expression of each InsP3 receptor subtype, empty Ca++ stores remain capable of activating a Ca++ entry pathway (Sugawara et al., 1997). Nevertheless, it remains important to establish the precise roles of the different InsP3 receptor subtypes (Mikoshiba, 1997) in regulating the cytosolic [Ca++]. However, establishment of the roles of InsP3 receptors in generating complex cytosolic Ca++ signals (DeLisle et al., 1996) and determination of whether InsP3 receptors are invariably essential elements of the pathways linking stimuli to physiological responses (Acharya et al., 1997) have been limited by the lack of both adequately selective antagonists of InsP3 receptors and of ligands that discriminate between receptor subtypes.
Adenophostin A is the most potent known agonist of type 1 InsP3 receptors (Takahashi et al., 1994a;Hirota et al., 1995), and we recently established that it is similarly potent in causing Ca++ release from the intracellular stores of permeabilized hepatocytes (Marchant et al., 1997a), in which type 2 InsP3 receptors are thought to predominate (Wojcikiewicz, 1995; De Smedt et al., 1997) The structure of adenophostin A (Fig. 1) suggests that its 3′′,4′′-bisphosphate structure with its adjacent 2′′-hydroxyl group may mimic the critical 4,5-bisphosphate/6-hydroxy triad of Ins(1,4,5)P3and related active analogs. The structures of, and abbreviations for, the adenophostin analogs are shown in Fig.1, and the abbreviations for the dissaccharide polyphosphates are given in the legend to Fig. 3. As the most potent known ligand of InsP3receptors that is not metabolized by the enzymes that degrade Ins(1,4,5)P3 (Takahashi et al., 1994a), adenophostin A provides a structure from which to attempt to devise novel ligands of InsP3 receptors. Such compounds may be more useful than the analogs of Ins(1,4,5)P3 that have hitherto provided the major source of ligands (Potter and Lampe, 1995). We recently demonstrated that an analog of adenophostin A [3-O-(α-d-glucopyranosyl)-β-d-ribofuranoside-2,3′, 4′-trisphosphate; RibP3](Fig. 1), which lacks the adenine moeity of adenophostin, is less active than adenophostin but as potent as Ins(1,4,5)P3 in causing Ca++ release from permeabilized hepatocytes (Marchant et al., 1997a). In view of this evidence suggesting that the adenine of adenophostin may play an important role in its high-affinity interaction with InsP3 receptors, we examined the behavior of a range of adenophostin A analogs (1-5) in which the adenine structure was preserved, while the ribose, glucose, and regiochemistry of the phosphate substituents have been altered (Fig. 1). In the present study, we establish that one of these compounds, (2S)-9-[1-(α-d-glucopyranosyl-3′′,4′′-bisphosphate)-2′-monophosphate-prop-3′-yl]adenine, which we have named acyclophostin (3), has unusual properties.
Experimental Procedures
Materials.
3H-Ins(1,4,5)P3 (58 Ci/mmol) was from Amersham (Little Chalfont, UK) and45CaCl2 was from ICN (Thame, UK). Ins(1,4,5)P3 was from American Radiolabeled Chemicals (St. Louis, MO). Adenophostin A, purified from Penicillium brevicompactum (Takahashi et al., 1994a), was a gift from Dr. M. Takahashi (Sankyo Co. Ltd., Japan). The disaccharide analogs of adenophostin (Marchant et al., 1997a), the analogs1-5 (Van Straten et al., 1997a, 1997c), and adenophostin A (Van Straten et al., 1997b) were synthesized and quantified as described previously. All ligands were analyzed by ion-exchange high performance liquid chromatography for isomer purity as fully described by Van Straten et al. (1997a, 1997b, 1997c). Percoll (1.13 g/ml) was from Pharmacia (Uppsala, Sweden). Ionomycin was from Calbiochem (Nottingham, UK), and thapsigargin was from Alamone Laboratories (Jerusalem, Israel). All other reagents were from suppliers listed previously (Marchant et al., 1997a).
Preparation of Rat Liver Membranes.
The liver of a male Wistar rat (200–250 g) was perfused in situ with 40 ml of ice-cold buffered saline [116 mM NaCl, 5.4 mM KCl, 0.96 mM NaH2PO4, 0.8 mM MgSO4, 25 mM NaHCO3, 1 mM EGTA, 11 mM glucose, 5% CO2/95% O2, pH 7.4, at 2°C]. After excision, the liver was chopped and then homogenized in 25 ml of ice-cold buffered sucrose [250 mM sucrose, 5 mM HEPES, 1 mM EGTA, pH 7.4, at 2°C] using a 15-ml glass Dounce homogenizer with 10 strokes of a loose-fitting plunger and 3 strokes with a tighter plunger. The homogenate was made up to 50 ml in ice-cold buffered sucrose, filtered through gauze, and centrifuged (2500g, 10 ml), and the pellet then was resuspended in 48 ml of ice-cold buffered sucrose containing Percoll (11.8% final v/v) (Prpic et al., 1984). The suspension was centrifuged (35,000g, 30 min), and membranes were harvested as a discrete fluffy band below a fatty layer at the top of each tube. The membranes were resuspended in 50 ml of ice-cold hypo-osmotic buffer (1 mM EGTA, 5 mM HEPES, pH 7.4, at 2°C) to lyse the vesicles and then centrifuged (48,000g, 10 min). The final membrane pellet was resuspended in binding medium (BM: 20 mM Tris, 1 mM EDTA, pH 8.3, at 2°C) at ∼20 mg protein/ml and stored in liquid nitrogen for up to 14 days. Protein concentrations were measured using the Bradford assay with bovine serum albumin as standard. A single liver typically provided ∼70 mg of membrane protein. Although this method produces membranes enriched in markers for plasma membrane, microsomal markers are also present (Prpic et al., 1984), and the membranes are enriched in InsP3 receptors whose characteristics have so far proved indistinguishable from those of permeabilized rat hepatocytes (Marshall and Taylor, 1994).
3H-Ins(1,4,5)P3 Binding.
Liver membranes (0.4 mg protein/tube) were added to BM (pH 7.0 or 8.3) (500 μl) containing 3H-Ins(1,4,5)P3 (30–60 nCi, final concentration, 1–2 nM) and the appropriate concentration of competing ligand. After 5 min at 2°C, bound and free3H-Ins(1,4,5)P3 were separated by centrifugation (20,000g, 5 min, 2°C). Previous results established that under these conditions, binding reached equilibrium and degradation of 3H-Ins(1,4,5)P3 was negligible. Total binding was typically 4000 dpm/tube, and nonspecific binding was approximately 30% of total binding.3H-Ins(1,4,5)P3 binding to rat cerebellar membranes was characterized as described previously (Richardson and Taylor, 1993).
45Ca++ Release from Permeabilized Rat Hepatocytes.
Hepatocytes were isolated by collagenase digestion of the livers of male Wistar rats (200–250 g) as described previously (Richardson and Taylor, 1993) and stored at 4°C in Eagle’s medium supplemented with 26 mM NaHCO3 and bovine serum albumin (2 g/100 ml) for up to 24 h. Cells were permeabilized by incubation with saponin (10 μg/ml) in a cytosol-like medium [CLM: 140 mM KCl, 20 mM NaCl, 1 mM MgCl2, 1 mM EGTA, 20 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES), pH 7.0, at 37°C] and subsequently loaded to steady-state (1–2 nmol Ca++/106 cells) by incubation (107 cells/ml) for 5 min at 37°C in CLM supplemented with CaCl2 (300 μM, [Ca++]c = 200 nM), ATP (7.5 mM), carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP; 10 μM), and 45Ca++ (7.5 μCi/ml). Unidirectional 45Ca++ efflux from the intracellular stores was initiated by diluting the cells (5-fold) into Ca++-containing CLM at 37°C supplemented with thapsigargin (final concentration, 1 μM). Appropriate concentrations of Ins(1,4,5)P3, adenophostin A or related compounds were then added, and 60 s later the 45Ca++content of the intracellular stores was determined after quenching in ice-cold medium (310 mM sucrose, 1 mM trisodium citrate) and then rapid filtration through Whatman GF/C filters using a Brandel receptor-binding harvester (Marshall and Taylor, 1994). A similar method was used for experiments in which 45Ca++release was measured under different conditions (e.g., pH , temperature) than those used to load the stores.
To compare 3H-Ins(1,4,5)P3binding and 45Ca++ release under identical conditions, permeabilized hepatocytes loaded with Ca++([Ca++]c = 200 nM; with 7.5 μCi of 45Ca++/ml for flux assays) were centrifuged (650g, 2 min) and resuspended in BM at 2°C supplemented with 140 mM KCl and 20 mM NaCl to give a final cell density of 2.5 × 106 cells/ml. Equilibrium binding of3H-Ins(1,4,5)P3 (3 nM) was measured after 5 min and unidirectional45Ca++ release was measured 3 min after the addition of either Ins(1,4,5)P3or acyclophostin.
45Ca++ Release from Cerebellar Microsomes.
Microsomes were prepared from rat cerebella as described previously (Patel et al., 1997). Briefly, cerebella from five male Wistar rats were homogenized in a Teflon-glass homogenizer in 9 volumes of ice-cold medium containing 250 mM sucrose, 5 mM HEPES (pH 7.05), 10 mM KCl, 1 mM MgCl2, and a cocktail of protease inhibitors (100 μM phenylmethylsulfonyl fluoride, 1 μM pepstatin, 0.02 unit/ml aprotinin, 20 μg/ml soybean trypsin inhibitor, 100 μM captopril). After two centrifugations (1000g, 5 min), the pellet was resuspended in homogenization medium and centrifuged again (9000g, 10 min). The supernatants from each of the three centrifugation steps were pooled and centrifuged (100,000g, 75 min), and the resulting microsomal pellet was resuspended in homogenization medium (∼15 mg protein/ml) and stored in liquid nitrogen. Microsomes (50 μl) were diluted into loading medium (2.73 ml) containing 100 mM KCl, 20 mM NaCl, 5 mM MgCl2, 20 mM HEPES (pH 7.0), 240 μM EGTA, 64 μM CaCl2 ([Ca++]c ∼200 nM), 1.5 mM ATP, 5 mM phosphocreatine, 1 unit/ml creatine phosphokinase, 8 μCi/ml45Ca++, and the protease inhibitor cocktail. After 5 min at 20°C, microsomes (20 μl) were added to appropriate agonists, and after 45 s, their 45Ca++contents were determined by stopping the incubations in ice-cold medium (100 mM KCl, 20 mM NaCl, 5 mM MgCl2, 20 mM HEPES, pH 7.0, 1 mM EGTA) and rapid filtration through Whatman GF/C filters (Nunn and Taylor, 1990).
Rapid Kinetics of 45Ca++ Release from Permeabilized Rat Hepatocytes.
Permeabilized hepatocytes loaded with 45Ca++ were immobilized between the filters of our superfusion apparatus, and media (20°C) were delivered to the cells (2 ml/s) from pressurized cylinders regulated by computer-controlled solenoid valves (Marchant et al., 1997b). The effluent containing the 45Ca++released from the cells was collected into vials (80 ms/fraction) arranged around the circumference of a programmable turntable. Details of the equipment and superfusion methods have been described previously (Marchant et al., 1997b). The equipment allows rapid (half-time = 46 ± 6 ms) exchange of the medium surrounding the permeabilized cells while measuring unidirectional 45Ca++efflux from them with high temporal resolution (80 ms) and under conditions where the composition of the medium (including its [Ca++]c) is rigorously controlled. At the end of each run, cells were superfused with CLM containing Triton X-100 (0.05%) to release all of the 45Ca++ remaining within the stores. Radioactivity (45Ca++ and the inert marker, 3H-inulin) within each sample was determined by liquid scintillation counting after the addition of EcoScint-A scintillation cocktail (National Diagnostics, Aylesbury, UK).
All traces were corrected for the unstimulated rate of45Ca++ efflux, and the amount of 45Ca++ released into each vial was then expressed as a fraction of the total45Ca++ content of the intracellular stores.
Analysis.
Equilibrium-competition binding curves were fitted to four-parameter logistic equations using a nonlinear curve-fitting program (Kaleidagraph, Synergy Software, PA)
Results and Discussion
Effects of Adenophostin A on Liver.
Maximally effective concentrations of Ins(1,4,5)P3 (10 μM) and adenophostin A (1 μM) released Ca++ from the same intracellular stores of permeabilized hepatocytes: alone or in combination they released ∼55% of the intracellular Ca++ pool. Adenophostin A, however, was significantly more potent than Ins(1,4,5)P3, with the half-maximal response to it (EC50 = 12.3 ± 0.3 nM) occurring at an ∼12-fold lower concentration than that to Ins(1,4,5)P3 (EC50 = 153 ± 11 nM) (Table1). In equilibrium competition binding studies to hepatic membranes, adenophostin A (K d = 1.60 ± 0.37 nM) bound to a single class of 3H-Ins(1,4,5)P3-binding site with substantially greater affinity than Ins(1,4,5)P3(K d = 7.96 ± 1.02 nM) (Table 1). We conclude, in keeping with results from other cells (Takahashi et al., 1994a; Hirota et al., 1995; Murphy et al., 1997; Missiaen et al., 1998), that in permeabilized hepatocytes, adenophostin A is the most potent agonist of InsP3 receptors yet identified. From both functional and radioligand binding analyses of liver and cerebellum, the properties of natural and totally synthetic adenophostin A (Van Straten et al., 1997b) were very similar (Table 1); the synthetic compound was used for most experiments. In subsequent experiments, we examined the effects of several structural analogs (Fig. 1; 1–5) of adenophostin A on hepatic InsP3 receptors.
Adenophostin A has within its structure an adenine moeity (Fig. 1) that is apparently required for optimal binding to InsP3 receptors (Marchant et al., 1997a), and because adenine nucleotides also bind to InsP3 receptors (Missiaen et al., 1997), the high-affinity binding of adenophostin A might have resulted from it simultaneously binding to the InsP3- and adenine nucleotide-binding sites. Our results suggest that such an explanation is unlikely because preincubation (3 min) of cerebellar membranes with adenine, adenosine, ADP, AMP, or adenosine-2′-monophosphate (1 mM) (Missiaen et al., 1997) did not selectively inhibit binding of adenophostin A relative to Ins(1,4,5)P3 (not shown).
Effects of Modified Adenophostin Analogs in Liver.
The effects of five modified analogs of adenophostin A (Fig. 1) on Ca++release from the intracellular stores of permeabilized hepatocytes are summarized in Table 1. Two of the analogs (2, 5) were inactive at concentrations of ≤100 μM. Three related analogs (1, 3, 4) were active, with each causing release of the entire Ins(1,4,5)P3-sensitive Ca++ store as revealed by the inability of a subsequent addition of Ins(1,4,5)P3 (10 μM) to cause further Ca++ release. The most potent of the analogs (3) (EC50 = 209 ± 12 nM), which we have named acyclophostin, was almost as potent as Ins(1,4,5)P3 (Table1). In equilibrium competition binding studies to hepatic membranes, four of the analogs (1-4) completely displaced specifically bound 3H-Ins(1,4,5)P3, although the affinity of 2 was extremely low (K d = 4.2 μM), in keeping with its lack of functional effect. Acyclophostin bound with substantially greater affinity (K d = 2.76 ± 0.26 nM) than Ins(1,4,5)P3 (K d = 7.96 ± 1.02 nM) and with only marginally lower affinity than adenophostin A (K d = 1.60 ± 0.37 nM) (Fig.2). Acyclophostin has the highest affinity for InsP3 receptors of any adenophostin analog with a modified ring structure yet synthesized.
There have been few attempts to define the structural determinants of the high-affinity interaction of adenophostin A with InsP3 receptors, although each active adenophostin analog has so far proved to be a full agonist. Most results are consistent with the 3′′,4′′-bisphosphate and 2′′-hydroxyl of adenophostin mimicking the essential 4,5-bisphosphate and 6-hydroxyl of Ins(1,4,5)P3 (Potter and Lampe, 1995) and with the 2′-phosphate of adenophostin mimicking the 1-phosphate of Ins(1,4,5)P3 (Fig. 1). The latter is supported by the observation that deletion of the 2′-phosphate from the ribose of adenophostin A massively reduced its binding affinity (Takahashi et al., 1994b). Glc(2′,3,4)P3 (Fig. 1), which was recently synthesized by two groups (Wilcox et al., 1995; Jenkins and Potter, 1996), is a low affinity agonist of InsP3 receptors (Marchant et al., 1997a), but its affinity is increased to almost match that of Ins(1,4,5)P3 when the flexible side chain to which the 2′-phosphate is attached is conformationally restricted in an anhydroerythritol derivative (Tatani et al., 1998).
Conformational restriction and phosphate positioning were also shown to be important in our previous study of the activity of disaccharide polyphosphates related to adenophostin (Marchant et al., 1997a). Ribophostin (RibP3), for example, essentially is adenophostin A without an adenine group (Fig. 1) and is as potent as Ins(1,4,5)P3 in evoking Ca++ release; until the present work, it represented the most potent synthetic analog known (Marchant et al., 1997a). Conformational restriction together with a nucleoside-type base therefore seems to be essential for analogs to be more potent than Ins(1,4,5)P3. The analogs used in the present study were designed to further explore the determinants of the increased potency of adenophostin by introducing modifications in the ribose and glucose rings and the regiochemistry of phosphate substitution.
It is not surprising that 5 was inactive: opening of the cyclic structure of the glucose-related ring presumably allows too much conformational mobility around the essential vicinal bisphosphate and neighboring hydroxyl group, which are essential elements of the pharmacophore (Potter and Lampe, 1995). A similar acyclic analog of Ins(1,4,5)P3 had previously been shown to be inactive (unpublished observation).
The substantial decrease (>200-fold) in the affinity of 1for InsP3 receptors parallels the massive decrease in the affinity of adenophostin A after removal of its 2′-phosphate (Takahashi et al., 1994b) and of Ins(1,4,5)P3 after removal of its 1-phosphate (Potter and Lampe, 1995). These results suggest that a nucleoside alone, in this case an acyclo-type nucleoside, does little to augment the activity of the glucose bisphosphate motif of adenophostin. The addition of a phosphate group to the 5′′-position of the glucose ring of 1 to give 2 further reduces biological activity. There is an inexact parallel with the much lesser affinity of Ins(1,3,4,5)P4 relative to Ins(1,4,5)P3 for InsP3 receptors (Burford et al., 1997), part of which may result from the additional 3-phosphate completely altering the ionization behavior of the 4,5-bisphosphate (Guédat et al., 1997).
Compound 4 appeared to be a relatively potent agonist, being only 10- to 20-fold less potent than Ins(1,4,5)P3. This was surprising because every known agonist of Ins(1,4,5)P3 receptors has a vicinal bisphosphate motif, but there is no such motif in 4(Fig. 1). A possible explanation might be to suggest that the 2′- and 2′′-phosphates of 4 may be close enough together to mimic the vicinal 4,5-bisphosphate of Ins(1,4,5)P3. This would suggest a novel orientation within the receptor in which the 4′′-phosphate and 3′′-hydroxyl of 4 might correspond to the 1-phosphate and 6-hydroxyl of Ins(1,4,5)P3. We cannot, however, exclude the possibility that migration of a protecting group during a prephosphorylation stage in the synthesis of4 may have lead to minor contamination with adenophostin A. It is difficult to eliminate such an explanation because it would require <1% contamination of 4 with adenophostin A to account for its observed activity, and such contamination cannot be excluded by standard spectroscopic and analytical techniques.
Acyclophostin (3) was almost as potent as Ins(1,4,5)P3 and similar to ribophostin (EC50 = 213 nM) (Marchant et al., 1997a) in causing Ca++ release from hepatic Ca++ stores (Table 1) and significantly more potent than Glc(2′,3,4)P3(EC50 = 1867 nM) (Marchant et al., 1997a). The adenosine of acyclophostin is connected to the glucose bisphosphate via a short linker with an attached phosphate group, and the stereochemistry at this center is the same as at the 2′-phosphate of adenophostin. These results therefore suggest that conformational restriction via a second ring structure (as in ribophostin and adenophostin) is not the only way to improve the biological activity of adenophostin analogs relative to Glc(2′,3,4)P3. Presumably, both the 2′-phosphate and the adenine of acyclophostin contribute to its activity at InsP3receptors but rather less effectively than for adenophostin. Despite some attempts at molecular modeling of adenophostin analogs (Wilcox et al., 1995), the apparently flexible nature of acyclophostin presents considerable difficulties in identifying a single low-energy conformation, making it unlikely that modeling of acyclophostin alone will provide unambiguous information. This approach may become more practicable after synthesis and biological evaluation of related conformationally restricted ligands.
Efficacy of Acyclophostin in Liver and Cerebellum.
Acyclophostin bound to hepatic InsP3 receptors with appreciably higher affinity than Ins(1,4,5)P3, but it was marginally less potent in causing Ca++ mobilization (Table1), suggesting that acyclophostin might be a partial agonist. Although the radioligand binding and functional analyses were performed under different conditions to optimize the signals obtained from them, comparison of the EC50/K d ratio for each of the agonists provides an index of their relative efficacies. The EC50/K d ratio was similar for each of 11 different agonists, including adenophostin A and two of the active ribose-modified adenophostin analogs (1, 4), but the ratio for acyclophostin was significantly (p < .05) higher (by ∼4-fold) than that for any other agonist (Fig. 3). The results suggest that acyclophostin may be a partial agonist of InsP3 receptors. A similar comparison of EC50/K d ratios, although with binding and functional analyses performed under different conditions and with different cell types, was previously used to suggest that Ins(1,2,3,5)P4 might be a partial agonist of InsP3 receptors (Burford et al., 1997).
An alternative explanation for the higher EC50/K d ratio for acyclophostin is that the permeabilized hepatocytes used for functional assays and the liver membranes used for radioligand binding might express different InsP3 receptor subtypes with different relative affinities for acyclophostin and Ins(1,4,5)P3. Both reverse transcription polymerase chain reaction analysis (De Smedt et al., 1997) and subtype-selective antibodies (Wojcikiewicz, 1995) of whole liver homogenates concur in suggesting that >80% of the InsP3 receptors are type 2 and the remainder are type 1. However, the exact distribution of subtypes is unknown within either our hepatocyte preparation (>95% hepatocytes) or the hepatic membranes prepared from whole liver (in which ∼65% of cells are hepatocytes). Unfortunately, with the InsP3 receptor subtype-selective antibodies currently available and the relatively low density of InsP3 receptors in permeabilized hepatocytes, we were unable to directly compare levels of expression of receptor subtypes in the two preparations. However, other evidence suggests that receptor heterogeneity is unlikely to be the cause of the difference in the EC50/K d ratio for acyclophostin and Ins(1,4,5)P3.
First, cerebellum expresses almost exclusively type 1 InsP3 receptors (Wojcikiewicz, 1995), yet under identical equilibrium competition binding conditions, cerebellar and hepatic membranes had very similar affinities for Ins(1,4,5)P3 and acyclophostin (Table 1). Acyclophostin is unlikely, therefore, to discriminate between the two InsP3 receptor subtypes (1 and 2) expressed in liver.
Second, when radioligand and functional assays were performed under identical conditions using permeabilized hepatocytes in Ca++-free medium at 2°C and pH 8.3 (see Experimental Procedures), the EC50/K d ratio for acyclophostin was again higher than that for Ins(1,4,5)P3 (Table2). Similar experiments with membranes (for radioligand binding) and microsomes (for Ca++ release) from cerebellum established that here, too, the EC50/K d ratio was higher for acyclophostin (not shown).
We conclude that acyclophostin does not discriminate between the InsP3 receptor subtypes (types 1 and 2) expressed in liver and that its high EC50/K d ratio therefore indicates that at pH 8.3 it behaves as a partial agonist of both hepatic and cerebellar InsP3receptors.
Acyclophostin Is a pH-Dependent Partial Agonist.
When the comparison of the effects of acyclophostin and Ins(1,4,5)P3on 45Ca++ mobilization from permeabilized hepatocytes and binding to hepatic membranes was performed at pH 7, there was no significant difference in the EC50/K d ratio for acyclophostin and Ins(1,4,5)P3 (Table 3). It was impracticable to perform both assays on permeabilized hepatocytes at pH 7 because the density of InsP3-binding sites was too low to permit their characterization at this suboptimal pH . However, because the relative affinities of acyclophostin and Ins(1,4,5)P3 for the receptors in permeabilized hepatocytes and hepatic membranes are indistinguishable at pH 8.3, we are confident that radioligand binding to hepatic membranes provides a valid comparison with permeabilized cells. These results therefore suggest that at pH 7, acyclophostin and Ins(1,4,5)P3 may be similarly efficacious in evoking Ca++ release from permeabilized hepatocytes. We conclude that when radioligand and functional assays are performed under similar conditions, acyclophostin appears to be a partial agonist at high pH and a full agonist at pH 7.
Rapid Kinetics of Ca++ Mobilization.
Most analogs of Ins(1,4,5)P3 are suggested to be full agonists (Potter and Lampe, 1995; Marchant et al., 1997b), but conventional measurements of the extent of Ca++ release are ill-suited to reliable measurement of efficacy. Even a partial agonist may be capable, as shown for acyclophostin (Fig. 2), of releasing the entire Ins(1,4,5)P3-sensitive Ca++ store but would be expected to do so more slowly (Safrany et al., 1993; Marchant et al., 1997b). The efficacy of agonists of InsP3 receptors can best be resolved by measuring initial rates of45Ca++ mobilization, which more closely reflect the extent to which InsP3 receptors have opened (Marchant et al., 1997b). In the final experiments, we used rapid superfusion of immobilized cells to determine the rates of45Ca++ release evoked by acyclophostin and Ins(1,4,5)P3. For practical reasons, our rapid superfusion experiments are restricted to 20°C, but because the discrepant EC50/K d ratio for acyclophostin is similar whether 45Ca++ release is measured at 2°C or 37°C (not shown), the need to perform the kinetics experiments at 20°C should not compromise their utility.
Figure 4A illustrates the typical response to a maximal concentration of Ins(1,4,5)P3 (5 μM): a rapid acceleration toward a peak rate of45Ca++ release is abruptly followed by a biphasic decay in the rate of45Ca++ release. The mechanisms underlying this pattern of Ins(1,4,5)P3-evoked Ca++release have been addressed previously (Marchant et al., 1997b;Marchant and Taylor, 1998), and the kinetic characteristics of the behavior observed in the present study were indistinguishable from those published previously. With the limited amounts of acyclophostin available, it was impracticable to complete such a detailed analysis, which would have required maximal stimulation with acyclophostin for several seconds. Instead, we examined only the initial responses to supramaximal concentrations of acyclophostin (25 μM) and Ins(1,4,5)P3 (5 μM). At pH 7, the responses to acyclophostin and Ins(1,4,5)P3 were indistinguishable: the time taken to reach the peak rate of45Ca++ release (320 ± 25 ms, n = 5; 293 ± 27 ms, n = 9, respectively), the peak rate of45Ca++ release (Fig. 4B), and the decaying phase of the response (not shown) were all similar. These results are entirely consistent with our suggestion that at pH 7, acyclophostin and Ins(1,4,5)P3 are each a full agonist of hepatic InsP3 receptors. At pH 8.3, however, the time taken to reach the peak rate of45Ca++ release was longer for acyclophostin (500 ± 30 ms, n = 4) than for Ins(1,4,5)P3 (420 ± 25 ms,n = 4), and the peak rate evoked by acyclophostin was only 69 ± 7% of that evoked by Ins(1,4,5)P3 (Fig. 4B). Analysis of the rapid kinetics of 45Ca++ release therefore confirms our suggestion that at pH 7, acyclophostin is a full agonist, but at pH 8.3, it is a partial agonist of InsP3 receptors.
Conclusions.
Most active inositol phosphate analogs appear to be full agonists of InsP3 receptors (Potter and Lampe, 1995), although it must be recognized that the assays used in most laboratories do not measure rates of Ca++ release and may thereby fail to detect a partial agonist unless it has very low intrinsic activity.
Several phosphorothioate analogs of InsP3(Safrany et al., 1993), including 3-deoxy-3-fluoro-inositol-1-phosphate 4,5-bisphosphorothioate (Wilcox et al., 1997) andd-myo-inositol-1,4,6-trisphosphorothioate (Mills et al., 1995), and perhaps some deoxy analogs (Mezna and Michelangeli, 1996), Ins(2,4,5)P3 (Marchant et al., 1997b), Ins(1,2,3,5)P4 (Burford et al., 1997), and Ins(1,3,4,6)P4 (Gawler et al., 1991) have all been reported to be partial agonists of InsP3 receptors. However, each of these partial agonists has a substantially lesser affinity than Ins(1,4,5)P3 for the receptor; even the highest affinity analog has 10-fold lower affinity than Ins(1,4,5)P3 (Wilcox et al., 1997). Acyclophostin is the highest affinity partial agonist of InsP3 receptors so far identified.
The interaction between acyclophostin and InsP3 receptors is unusual in that acyclophostin is a full agonist at pH 7 but a partial agonist at pH 8.3 (Fig. 4). Another analog, 3-amino-3-deoxy-Ins(1,4,5)P3, has also been reported to show pH-dependent efficacy: it is a full agonist at pH 7.2 but a partial agonist at pH 6.8 (Kozikowski et al., 1994). It is not, however, clear whether under the conditions used for those comparisons, the decrease in pH was not also accompanied by a substantial increase in free [Ca++], which would inhibit the type 1 InsP3 receptors of the SH-SY5Y cells used for the assays. We cannot yet resolve whether the pH-dependent ability of acyclophostin to activate InsP3receptors results from an effect of pH on the receptor or the ligand. The former would be intriguing because it would suggest that from among the many analogs examined, acyclophostin is uniquely dependent on specific pH-sensitive residues within the receptor to cause channel opening subsequent to its binding to the receptor.
Unfortunately, there have been only a few systematic studies of the pK a values of the phosphate groups of inositol phosphates; furthermore, values measured in aqueous media may differ substantially from those in less polar solvents (Tribolet and Sigel, 1987), which may more closely mimic the environment within the binding site of the receptor. The pK avalues for Ins(1,3,4,5)P4 and Ins(1,2,4,5)P4 have been determined (Guédat et al., 1997), as have those for a conformationally restricted InsP3 analog (Riley et al., 1998). In light of these data, the phosphate groups of adenophostin are unlikely to be fully ionized at pH 7 or pH 8.3, and by analogy with Ins(4,5)P2, the bisphosphate motif might be expected to have third and fourth pK avalues at about 6 and 8 (Schmitt et al., 1993) (i.e., within the pH range that affects the efficacy of acyclophostin). Although the 3-hydroxyl of Ins(1,4,5)P3 and the 5′′-hydroxymethyl group of acyclophostin may have different effects on the pK a of the neighboring (4 or 4′′)-phosphate, it is difficult to envisage large differences in the phosphate pK a values for adenophostin and acyclophostin. The conformational restriction imposed by the furanoside ring of adenophostin may influence the pK aof its 2′-phosphate, and that ring is, of course, missing from acyclophostin, where the 2′-phosphate may interact directly with the base (Fig. 1). In short, without directly determining the pK a values of acyclophostin and comparing them with the other analogs, it is impossible to eliminate the possibility that the pH-dependent efficacy of acyclophostin results from a conformational change in the ligand. An alternative explanation would be that although Ins(1,4,5)P3 and acyclophostin share substantially overlapping binding sites, they may differ in the receptor residues they contact to cause channel opening.
We conclude that acyclophostin has the highest affinity of any adenophostin analog with a modified ring structure so far synthesized and that its unusual pH-dependent efficacy may result from effects of pH on either the ligand or the receptor. The latter would suggest that different agonists may use different residues within a receptor to cause its activation.
Acknowledgments
We thank Dr. M. Takahashi for the generous gift of natural adenophostin A.
Footnotes
- Received July 14, 1998.
- Accepted September 25, 1998.
-
Send reprint requests to: Dr. Colin W. Taylor, Department of Pharmacology, Tennis Court Road, University of Cambridge, Cambridge, CB2 1QJ UK. E-mail: cwt1000{at}cam.ac.uk
-
This work was supported by grants from the Wellcome Trust to C.W.T. (039662) and B.V.L.P. (045491) and from the Biotechnology and Biological Sciences Research Council to C.W.T. J.S.M. is supported by a Wellcome Prize Fellowship (018484).
Abbreviations
- BM
- binding medium
- [Ca++]c
- medium-free Ca++concentration
- CLM
- cytosol-like medium
- EC50
- concentration causing half the maximal effect
- h
- Hill coefficient
- Ins(1
- 4,5)P3,d-myo-inositol-1,4,5-trisphosphate (other inositol phosphates are similarly abbreviated)
- InsP3
- inositol trisphosphate
- PIPES
- piperazine-N,N′-bis(2-ethanesulfonic acid)
- The American Society for Pharmacology and Experimental Therapeutics